Professional Documents
Culture Documents
Translating That: An Ideational Correspondence Analysis of Machine Translation
Translating That: An Ideational Correspondence Analysis of Machine Translation
1, April 2021
Received: 23 Januari 2020 | Accepted: 03 Februari 2021 | Published: 30 April 2021
TRANSLATING THAT:
AN IDEATIONAL CORRESPONDENCE ANALYSIS OF
MACHINE TRANSLATION
Ardik Ardianto1*
1
Universitas Gadjah Mada, Indonesia
*e-mail: ardikardianto@gmail.com
Abstrak
Penelitian ini bertujuan untuk membandingkan hasil terjemahan dua mesin terjemahan, yakni Google
Translate dan Bing Translator, dalam menerjemahkan leksem that pada artikel berita. Pendekatan yang
digunakan dalam menelaah kesejajaran hasil terjemahan dari leksem that pada penelitian ini adalah
linguistik sistemik fungsional, khususnya pada struktur eksperiensial dan logis. Jenis analisis yang
digunakan adalah komparatif deskriptif. Data yang digunakan meliputi 40 satuan kebahasaan dari leksem
that yang diambil dari enam artikel Berita Dunia BBC. Hasil temuan menunjukkan bahwa dua mesin
terjemahan mampu mengenali tiga fungsi leksem that, yakni fungsi inti (Head), fungsi post-modifier, dan
fungsi konjungsi. Fungsi terbanyak yang ditemukan adalah fungsi post-modifier (19 kali atau 47,5%), yang
diikuti dengan fungsi konjungsi (17 kali atau 42,5% pada mesin pertama dan 18 kali atau 45% pada mesin
kedua). Fungsi inti menjadi fungsi terakhir setelah kedua fungsi lainnya (empat kali atau 10% pada mesin
pertama dan tiga kali atau 7,5% pada mesin kedua). Lebih lanjut, dikarenakan adanya variasi elipsis dari
leksem that sebagai pronomina relatif dan post-determiner pada mesin terjemahan pertama, penelitian ini
menyimpulkan bahwa hasil terjemahan mesin pertama lebih akurat, berterima, kreatif, dan kontekstual
daripada hasil terjemahan mesin kedua.
Abstract
This paper aims to contrast the translation of two machine translation systems, Google Translate and Bing
Translator, in translating the lexeme in news articles. The approach used in scrutinizing the lexeme's
translation correspondence in this study is systemic functional linguistics, especially in both experiential
and logical structures. This study was carried out through descriptive comparative analysis. This study's
data were 40 constituents that were taken from six BBC World news articles randomly selected. A
thorough analysis demonstrates that the two machine translation systems can recognize the three
functions of that, i.e., Head, post-modifier, and conjunction. The highest emerging function is post-modifier
by 19 times (47.5%), followed by the conjunction function by 17 times (42.5%) on the first machine
translation system and 18 times (45%) on the second one. The lowest emerging function is Head by four
times (10%) on the first machine translation system and three times (7.5%) on the second one.
Furthermore, due to the elliptical variation of that as a relative pronoun and the translation variation of that
as a post-determiner, it concludes that the translation outputs of Google Translate are more accurate,
semantically acceptable, creative, and contextual than those of Bing Translator.
This study aimed at scrutinizing the Bing Translator. These two systems were
ideational structures of the lexeme that in selected on the grounds that both are the
the two machine translation systems. The most popular machine translation systems
analysis was carried out by contrasting, that are frequently used worldwide (Precup-
therefore finding out the similarities and Stiegelbauer, 2013). The data were
differences, the experiential structure and retrieved from the global news channel, that
logical structure of those lexemes. It was is BBC News, consisting of six news
specifically focused on two aspects, i.e., articles randomly selected. The data of this
function and category of the lexeme that. study comprised forty constituents that in
This study, moreover, was expected to be English (Source Text) and the translation in
of beneficiary for describing how the Indonesian language (Target Text). The
lexeme that was translated into the two data analysis was carried out in two stages:
machine translation systems and to what 1) the percentage calculation of each
extent the translation outputs were function and category of the lexeme that in
corresponded to the originals in terms of both machine translation systems; 2) the
both function and category in the elaboration of contrastive analysis
experiential and logical structures. (similarities and differences) of the
translation.
METHODS
This study is a contrastive analysis FINDINGS AND DISCUSSION
focusing on two major axes, i.e., similarities Based on the translation of the two
and differences of the translation of two machine translation systems, i.e., Google
machine translation systems in translating Translate (GT) and Bing Translator (BT),
the lexeme that. The machine translation each function and category of the lexeme
systems used were Google Translate and that is thoroughly elaborated in the Table 6.
Head 3 (7.5%)
pronoun (itu) 1 (2.5%)
adverb (sebelum itu) 1 (2.5%)
phrase (seperti itu) 1 (2.5%)
phrase (dari itu) 1 (2.5%) 19
Post-modifier
pronoun (yang) 10 (25%) (47.5%)
pronoun (yang satu) 1 (2.5%)
determiner (itu) 5 (12.5%)
Conjunction particle (bahwa) 18 (45%)
Total 40 (100%)
From the table 6 and 7 above, it and the deictic determiner itu (that) by five
can be seen that the most frequently times (12.5%). The difference of these two
occurring function of the lexeme that was machines relied on the adverbs
as post-modifier (19 times or 47.5% in each sebelumnya – sebelum itu (previously and
machine translation system). On the first prior to that) and relative pronouns yang itu
machine, such function of post-modifier was – yang satu (both mean that one).
distributed in the form of the adverb Furthermore, the second most occurring
sebelumnya (previously) once (2.5%), the function was as conjunction, realized in the
phrase seperti itu (such that) once (2.5%), form of the Indonesian particle bahwa
the phrase dari itu (of which) once (2.5%), (that). On the first machine, this function of
the relative pronoun yang (that) by ten conjunction occurred 17 times (42.5%)
times (25%), the demonstrative relative distributed into two, i.e., the Indonesian
pronoun yang itu (that one) once (2.5%), particle bahwa (that) by 15 times (37.5%)
and the deictic determiner itu (that) by five and Ø twice (5%); whereas, on the second
times (12.5%). Meanwhile on the second machine, it occurred 18 times (45%). The
machine, it was distributed in the form of function of Head occurred in the third place,
the adverb sebelum itu (prior to that) once in the form of the pronoun begitulah (that
(2.5%), the phrase seperti itu (such that) so) and itulah (that) by four times (10%) in
once (2.5%), the phrase dari itu (of which) the first machine and in the form of the
once (2.5%), the relative pronoun yang pronoun itulah (that) by three times (7.5%)
(that) by ten times (25%), the relative in the second one.
pronoun yang satu (that one) once (2.5%),